jpistell
123 posts
msg #33946
- Ignore jpistell |
11/15/2004 5:26:08 PM
Best Filter Contest.
Challenge Summary: Can your long term filter beat the Market?
Requirements:
Filter must produce at least 20 matching stocks
Stocks > $4
Avg Vol > 50k
Date SetBack: Anywhere from 75 to 90 days (July 8 to July 29, 2004)
Contest Goal:
Dial in your filter over the range of dates above. After everyone posts their filters, we'll have someone call a specific setback date and see who wins.
Contest Measurement:
We'll use the Filter Performance Window for results. The winner will have the highest CURRENT OVERALL % returns (as seen in the Overall column).
This setback will make for a difficult contest. The Nasdaq is up 7% for this period and the DOW is + 3%.
"May the Filter Be With You."
Enjoy,
Joe
p.s. more matching stocks = smoother, more predictable output throughout the setback range.
p.p.s. Only post one filter, so make it a good one!
|
yepher
359 posts
msg #33950
- Ignore yepher |
11/15/2004 8:36:18 PM
I don't understand why you would have the requirement:
"Filter must produce at least 20 matching stocks"
Who cares how many results are returned? Some days there maybe 1 while other days there maybe 50.
|
jpistell
123 posts
msg #33953
- Ignore jpistell |
11/15/2004 9:36:17 PM
Ideally, I thought of your idea. I really like testing filters in good markets and bad. I originally wanted to make it one filter competing on multiple setback dates. SF has reporting limitations that we need to work within. The contest needs to take place on ONE day to simplify the measurement.
But, why I am fishing for >20 hits? This requires the filter writer to produce a loose filter that produces the strongest uptrend they can configure. This filter isn't a razor sharp RSI2 tool, but a foundation of a trend. Take for instance The Rumpl'd One's 60day slope on the RSI2. The 60day slope is the foundation, the RSI2 is the trigger. This contest helps establish the foundation for all of us to profit from.
Joe
|
jpistell
123 posts
msg #33954
- Ignore jpistell |
11/15/2004 10:06:50 PM
Reference: This is almost "marked to market"
What ever the market does, so goes this filter. No foundation here.
|
jpistell
123 posts
msg #33955
- Ignore jpistell |
11/15/2004 10:14:35 PM
Now this is nearly 2:1 over the market from 75 to 95 days back.
This is my entry.
Joe
|
txtrapper
548 posts
msg #33956
- Ignore txtrapper |
11/15/2004 11:39:27 PM
My filter contest entry;
TxTrapper
--------------
|
yepher
359 posts
msg #33957
- Ignore yepher |
11/16/2004 12:23:49 AM
jpistell,
This is not my entry but you might try this to improve your filter results for that time span:
|
jpistell
123 posts
msg #33961
- Ignore jpistell |
11/16/2004 9:11:32 AM
TxTrapper,
Your filter is too tight. Yours is a trigger filter. It only produces 0-3 "hits". You need to loosen it up to capture > 20 hits. We're trying to capture trends not entry points.
Just as an example, If we were to loosen the RSI to 20, then you capture the min # needed, but, the filter has "lost its edge" because its designed as a trigger filter (i.e. capture a rare over-sold occurance without concern to longer term trends.)
Joe
|
jpistell
123 posts
msg #33962
- Ignore jpistell |
11/16/2004 9:15:19 AM
yepher,
lrs... very interesting indeed!
http://www.stockfetcher.com/stockdb/fetcher?p=forum&sub=view&fid=1006&tid=91
I will be tinkering with this tool tonight!
Many thnx!
Joe
|
yepher
359 posts
msg #33963
- Ignore yepher |
11/16/2004 11:11:50 AM
Ok here is the best I can do ;) Sorry it does not hit the 20 mark either but it does work every day for that range.
If I have to get 20 the best I can do is 80% every day.
|